The Design And Implementation Of Embeded P2P Streaming Media Broadcasting System Based On Peercast Model

With the rapid development of internet technology, P2P streaming network technology is more and more widely applied. The traditional streaming media mainly adopts C/S mode, servers establish connections with the clients in unicast way. Most of date provided by servers is streaming media data, for which the requirement of network bandwidth is higher and connection duration is longer. With the continuously increasing of the clients, the bandwidth and system resource would be exhausted in a short time. The performance and bandwidth will be the bottleneck of the whole system. Therefore, the research of P2P technology may have important implications in alleviating the pressure of streaming media servers.This thesis analyses the PeerCast model based on P2P technology, researches the system architecture and related algorithms of source code, and points out the defects existing in the system. PeerCast model node selection mechanism adopts depth-first algorithm, the node selection mechanism is primarily based on the depth of the node in multicast tree, but the depth of the node only represents the logic distance which can not fully reflect the distance between the nodes in physical topology of the underlying network, and this may lead to long-time delay for the downstream nodes of the muliticast tree.In view of the problems above, this thesis proposes a multicast tree balancing algorithm, of which the node selection strategy adopts depth-first in intranet and route-first in internet, and improves the node selection algorithm for PeerCast model. The multicast tree balancing algorithm tries to match node distance between overlay network and physical topology network. Finally we proves the feasibility of algorithm through effective experiments, reduces the delay of the downstream nodes of the multicast tree, and improves the real-time performances of the PeerCast system.